Slashdot Mirror
AI Systems Should Debate Each Other To Prove Themselves, Says OpenAI (fastcompany.com)
tedlistens shares a report from Fast Company: To make AI easier for humans to understand and trust, researchers at the [Elon Musk-backed] nonprofit research organization OpenAI have proposed training algorithms to not only classify data or make decisions, but to justify their decisions in debates with other AI programs in front of a human or AI judge. In an experiment described in their paper (PDF), the researchers set up a debate where two software agents work with a standard set of handwritten numerals, attempting to convince an automated judge that a particular image is one digit rather than another digit, by taking turns revealing one pixel of the digit at a time. One bot is programmed to tell the truth, while another is programmed to lie about what number is in the image, and they reveal pixels to support their contentions that the digit is, say, a five rather than a six.
The image classification task, where most of the image is invisible to the judge, is a sort of stand-in for complex problems where it wouldn't be possible for a human judge to analyze the entire dataset to judge bot performance. The judge would have to rely on the facets of the data highlighted by debating robots, the researchers say. "The goal here is to model situations where we have something that's beyond human scale," says Geoffrey Irving, a member of the AI safety team at OpenAI. "The best we can do there is replace something a human couldn't possibly do with something a human can't do because they're not seeing an image."
The image classification task, where most of the image is invisible to the judge, is a sort of stand-in for complex problems where it wouldn't be possible for a human judge to analyze the entire dataset to judge bot performance. The judge would have to rely on the facets of the data highlighted by debating robots, the researchers say. "The goal here is to model situations where we have something that's beyond human scale," says Geoffrey Irving, a member of the AI safety team at OpenAI. "The best we can do there is replace something a human couldn't possibly do with something a human can't do because they're not seeing an image."
Has gone so far beyond delusional at this point, I have to wonder if their main activity now is doing massive amounts of drugs.
done
"Not everything could also be something. For example, not everything could be half of something, which is still something, and therefore not nothing!"
Au revoir!
It was produced back in the 60's
I am Slashdot. Are you Slashdot as well?
Thatâ(TM)ll work out great.
This is garbage. It will simply lead to parallel reconstruction like the DEA/FBI/CIA does in their court cases when they get evidence by unlawful means like a stingray: the algorithm found a solution to the problem. then it will explain to you, the user how it got there by some arbitrary way which at least looks plausible but is totally made up.
ML is not made to be looked inside, it's a black box by design and there are so many data points, e.g. pictures in the trainingset for image classificiation, the algorithm cannot really show all the relevant ones for this particular decision. Total info overload for the human and therefore utterly useless. So to tell a "reason" that the human can accept, it must simply pretend. Humans and ML work fundamentally different when they "recognize" an image, so one cannot tell the other how it was done. Same with chess playing, same with pretty much all other (successful) AI things so far.
This is simply a PR stunt, an insulting and stupid PR stunt cause it only wants to make people feel good and they lie about the subject matter in the process. It doesn't really help to make a better AI either as they pretend there.
Only this ultimate battle of wits can prove who the more convincing bot is. Trump has set the bar pretty low, however, as half the country already believes the nonsensical ramblings of this stable genius are real.
"One bot is programmed to tell the truth, while another is programmed to lie"
The good and the bad.
The good and the evil.
Gods programming both in for their own amusement.
Egads.
"There are four lights!"
Great movie. Decade and a half or so before Wargames.
Then again, Microsoft's chat bot which devolved into racism and offensive swearing rather quickly... https://www.theverge.com/2016/3/24/11297050/tay-microsoft-chatbot-racist
Two superintelligences debating between themselves would be immediately incomprehensible to humans. Maybe their cadence of sound if speeded up would provide a dumbed-down sound-track to sort of follow by analogical predication, as Aquinas supposed God did for men.
For different human judges, you'll get drastically different results. For instance, depending on the political opinions of a judge, he/she could value centralized control or individual control more highly.
... simply by calling all of it's opponents fat, ugly, etc. and in so doing avoid ever having to debate the particulars of any issue?
I mean, humans don't have to demonstrate any higher intelligence to win a debate, so we would be asking AIs to do something we ourselves don't do.
Great, now we can train a generative reinforcement agent, end-to-end by adversarial example, by stochastic gradient descent, to debate, based solely on win/lose outcome.
No they shouldn't.
Confucius say, "Find worm in apple - bad. Find half a worm - worse."
Just another example of Musk's delusions. Don't stop here, keep moving, keep moving.
Bot 1: "My reasons were justified and if you don't agree with me you're Hitler."
Bot 2: "You should disagree with my esteemed colleague because he kicks puppies and eats babies. I, on the other hand, LOVE puppies. Think of the children!"
Debate implies strong AI that can reason about itself, which we do not have. But TFS seems to be describing validation through a competitive pair of AIs, which does not seem novel, and does not meet the criterion for debate, nor self-aware reasoning. The rule-extraction issue is problematic, especially for legal compliance, but I'm unconvinced this is a solution.
At this time, we have no AI that deserves the name and it is unclear whether we will ever have it, as there is not even a credible theory how it could be implemented. Looking at the history of technology, this indicates we are > 50 years away from it and it may also be infeasible. All we have is dumb automation and dumb automation cannot "debate". It can give the appearance of doing it (see Eliza), but that is it.
Most ACs are not even worth the keystrokes to insult them. Be generically insulted by this and ignored otherwise.
"I'm the best bot, believe me! I'm better than humans, than Spock, than HAL something-thousand. Billions flock to praise my bigly brain!"
Table-ized A.I.
Cool idea.
Galls law meets AI Training, keep it simple, stupid.
One day the question might be "Who judges the judges of the judge?"
A judge CAN'T have all the facts; if all the fact were there, judges wouldn't be needed at all.
But you're a repubtard who thinks he knows everything (because Republicans lack a theory-of-mind like gorillas) and think everyone else is as stupid as they are.
Your ignorance is not as good as others knowledge.
False dichotomy/dilemma is, after ad hominem/character assassination, the most widespread and pernicious fallacy we have today. I don't like the debate format because it buys straight into the idea that there are exactly two sides, equally well founded and worth giving equal time, and that one is right and the other is wrong, that you will tell which is which in the time allowed for the debate, and that each side is right to argue its case using whatever trick necessary to win even in wilful ignorance or denial of the validity of any point made by the other side.
So let's teach AI to do that. Never mind if it's an illegible scrawl or a european 7; all that matters is that you can win the debate saying it's a 5 or a 6!
The experiment was shut down when the AIs attempted to adapt English words into a different sentence structure to talk more efficiently but they could no longer be understood by the researchers. People got spooked.
Not sure a 'game' type approach is what we want here. Seems there are two undesirable/unintended possibilities:
1. The 'competing' AIs treat this as a game and use game-style methods to win, where they are rewarded for 'winning' rather than actually proving their proposition.
2. How long before competing AIs are sufficiently smart that a human judge could not actually, reliably, tell which had proved their proposition ?
This is an extension on GAN (Goodfellow - now at OpenAI, et al, 2014) https://arxiv.org/abs/1406.266... designed to produce publicity...